I've been surprised to read some comments and reviews about Turangalila over the last few days. I had assumed it was more or less part of the canon now. But clearly people have doubts of various kinds: the unevenness of the work, the 'cliched' melodies (even an 'Ice Cream Van Jingle'!!) the use of the ondes Martenot, the alleged self-indulgence of the composer, etc etc.

In my view it succeeds completely in what it sets out to do- yes it is uneven, but the languid passages, double-figure movement structure, the sudden orchestral eruptions, the massive climaxes, the obsinately memorable, 'simple' themes are all of a piece with that Messiaen was aiming for. At the risk of sounding trivial, one might ask the naysayers to 'lighten up'! 

Rattle held it together very well indeed, and it was light years away from his dull, over-cautious studio recording of the 1980s. One can only hope for a live recording this time. One thing Rattle shares with Leonard Bernstein (and isn't it a pity he never recorded it?) is that he seems to perform much more effectively in concert than in the studio. I can only imagine the more sterile environment of the studio brings out an overfussiness with detail and a consequent loss of overall vision. That certainly was not the case on Tuesday night.

There is documentation of Bernstein in Turangalîla. A half-hour rehearsal was recorded/broadcast by NBC on 28 November 1949, prefaced by an amazing little speech by Koussevitzky. Bernstein rehearses bits of the first, fifth and sixth movements (part of the latter was published in the Boston Symphony box, as you say). Sadly, the concert itself wasn't broadcast (it was one of the Boston seasons where they had no deal with the broadcasters). If you'd like to know more I could email you some information about all this...